Among the 12 specimens of Atretium schistosum available in the MNHN collections, only two females (16.6%) show parietal pits (MNHN 612 and MNHN 1946.64) (Table 1). One specimen (MNHN 612) presents two pits on the left parietal plate, meeting at the posterior part, forming a Y-shaped structure. The right parietal plate of this specimen is devoid of pits (Fig. 1b and d). The second specimen (MNHN 1946.64) shows a pit in the centre of each parietal plate (Fig. 1a and c). De Haan [1] mentioned the existence of small vesicles present along the bottom of the parietal pit. These are visible when the superficial level of the old epidermis is removed. These structures were not visible because the Paris specimens are relatively old and their state of preservation has not conserved the fragile parts. The similarity of the parietal pits observed in Atretium schistosum with comparable structures mentioned by de Haan [1] is evident in their shape, location, size and number. Such similarity appears to be homologous between these structures, suggesting possible inheritance from a common ancestor. De Haan [1] also noted that these parietal pits can occur symmetrically (47% of the specimens) or asymmetrically (53%); the specimens of A. schistosum examined here are in accordance with this observation.

Out of 10 of the 12 examined specimens, other structures were observed, including small protuberances located on the chin, at the level of the inferior labial scales and the mental plates, gular plates and symphysials, but also on some of the first ventral plates (Fig. 2, Table 1). Such structures are also encountered in some psammophine snakes, as noted by de Haan [6]. The simultaneous occurrence of both these structures in the same species seems to argue in favour of them being functionally linked. De Haan (op. cit.) suggested that they could have a function in chemical olfactory marking.

The exact position of the genus Atretium inside colubroid snakes is still debated, this genus having never been included in any phylogenetical analysis. According to Underwood (pers. commun. in [4]), the genus should be classified in the subfamily Xenodontinae. In the light of recent data [7,8], such a hypothesis is untenable and can reasonably be rejected, phylogenetic relationships of that subfamily showing that it is restricted to the New World. Two other subfamilies were suggested to include Atretium, the Natricinae, as suggested by Deuve [3], or the Colubrinae, as proposed by De Silva [4]. The last hypothesis, never formulated until now, would consist in including Atretium inside, or close to the psammophine, but given the high morphological, ecological and biogeographical differences between those two groups, this possibility appears to be very few probable. So, in the light of our observations, the occurrence of parietal pits could no more be taken as a unique synapomorphy of the Psammophiinae as suggested by de Haan [1]. These structures, probably glandular or sensory, should rather be considered as regressed organs constituting a plesiomorphy of the wider snake group of colubroids. Indeed, given that psammophine snakes are more closely related to the elapids and the atractaspidids than to most of the other colubrids [9], these pits could have been present at the base of the colubroid lineage and then regressed in most groups. Such a hypothesis is valid, whether Atretium is a natricine or a colubrine.

Two plausible hypotheses are here considered in order to account for the sporadic occurrence of parietal pits as well in Psammophiinae than in A. schistosum.

(1) It could be an atavistic phenomenon. According to Hall [10], atavisms are the reappearance of ancestral characteristics in individual members of a species. They constitute proof that the genetic and developmental information originally used in the production of such characteristics has not been lost during evolution, but lies quiescent within the genome and in the processes of embryonic development. This occasionally resurgent trait may or may not be functionally expressed. By way of examples, the domestic horse, Equus caballus, can occasionally possess supernumerary digits [11], and the development of teeth in chick embryos can be experimentally induced by transplanting mouse neural crest cells [12]. In colubroid snakes, parietal pits could occasionally ‘reappear’ in some individuals (as could be the case in Atretium and among some Psammophiinae); some snake lineages would be more likely to produce such atavistic individuals. The low number of occurrences of pits within one species seems to indicate that they no longer have a functional role, as is often the case for several complex atavistic organs. Caution should be exercised, however, when interpreting the occurrence of these pits in only two colubroid lineages. These could in fact exist in other lineages, but they may be scarcely visible, and thus have not attracted the attention of herpetologists. A similar example is provided by the recent discovery of an external epidermoid supralabial gland present in all species of the viperid genus Echis [13,14]. These snakes have been previously examined by many herpetologists, yet the gland not noticed, although the gland aperture is visible without magnification. That is why it would be interesting to search for the presence of parietal pits in the most basal lineages of Alethinophidians (homalopsids, viperids, xenodermatids, boids, etc.) in order to determine whether their occurrence is more widespread in the basal clades of snakes.

(2) The second hypothesis considers these structures as permanent organs by colubroid snakes at least. They could be present permanently but in a discreet way, hidden by the tegument. Only after an injury, or an anomaly in the embryonic development, or even during some pathologies, will these structures open to the exterior. Those parietal pits could then play an unsuspected secretory or sensorial role.

Smith [15] described the presence of nuchal glands in 11 species of Asian natricine snakes of the genus Balanophis M. Smith, 1938, Macropisthodon Boulenger, 1839 and Rhabdophis Fitzinger, 1843. Those dorsal paired glands are distributed in the neck (and could also extend to the whole length of the body in some species). They could exude an irritant secretion in order to repulse potential predators. However, contrary to the parietal pits, those organs have never been observed directly on the head of any specimen.

Concerning the presence of gular pits, de Haan [1] formulated several hypotheses regarding their function. One of these considers the pits as ‘contact joints’, which allow the snake to shed correctly, in retaining the skin to shed until the most favourable moment for shedding. Another hypothesis suggests that pits could play a role in sexual recognition during precopulatory parades and during copulation itself. Indeed, among many snakes, the male lays his head flat on that of the female during the copulation parade (Fig. 3). The supposed glandular gular pores, presumably secretory structures, are then in contact with the parietal pits of the female, which perhaps receive chemical signals or sensory information. This hypothesis seems less probable, because the frequency of occurrence of parietal pits is low inside the one species. Moreover, the hypothesis of parietal pits hidden under the parietal plate is incompatible with the transmission of chemical information from the gular pits of one individual to the receptor organs located in the parietal pits of another.